Zimex

#Problem With the rapid spread of urbanization in the world, both the number of permanent residents in cities and the population density are increasing. When a fire occurs, it seriously threatens people’s lives and causes major economic losses. To prevent fires, it is necessary to establish a monitoring system that can detect fires. There are various sensors for fire detection, including smoke alarms, temperature alarms, and infrared ray alarms. Although these can do the job, they have major flaws with those monitoring systems. 1.None of those can accurately identify the early stages of the fire. Therefore even the system identifies the fire, It may already be too strong to control. that can lead to huge disasters, leading to human, ecological and economic losses. 2.Those systems are very costly when it comes to applying them to large facilities. therefore major companies with high budgets can only afford them. 3.Most of the alarms can only be functional in a closed environment, which is ineffective for a wide space, such as outdoors or public spaces. #Solution Since many organizations, houses, and shops have CCTV systems, Establishing a real-time automatic fire monitoring algorithm for existing CCTV’s will be the solution for all those major issues. Using the power of Machine Learning, the algorithm can be trained to identify the early stages of the fire. which will perform better than any existing systems with regards to identifying the early fire. Since the algorithm is based on image-recognition, Algorithm can identify indoor fire as well as outdoor fire. which will hard to using typical fire detection systems. #The product To embed the algorithm into the existing CCTVs can be done by introducing a new IoT friendly Smart NVR (Network Video Recorder) device. Users just need to replace their old DVR(Digital video Recorder)/NVR with our new Smart NVR. Which will cost-effective rather than set up a huge fire detection system. #Technology Overview A raspberryPI with a light Operating system (Raspbian Stretch Lite) will be used as a NVR (Network Video Recording) hardware solution, as Raspberry has a H264 hardware encoding/decoding capabilities Shinobi is an Open Source CCTV solution that will install on that raspberry PI device and use it as a NVR software solution. It is developed with nodeJs therefore we can easily modify it according to our requirement. Using deep learning (customized InceptionV3 and CNN architectures) and OpenCV fire detection system will be trained.

Team Consist of two members. #Members 1.Dilan Perera Undergraduate of University of Colombo School of Computing Email: r.dilanperera@gmmail.com Phone: +94712479175 2.Nimaya Perera Undergraduate of University of Colombo School of Computing Email: nimayamanthi452@gmail.com Phone: +94764503326